Robert P

That's why I asked if anyone can offer a way to demonstrate it - i.e. see for myself that it happens.

joejack951

I imagine that with a fairly accurate caliper and a bare hub one could measure the length of the axle before and after applying a quick release (using something similar to open dropouts to load the hub leaving space to get at the axle for the measurement) and note the change in length. It won't be much, but should be in the repeatably measurable realm i.e. more than 0.02mm. I believe I calculated previously that a quick release skewer stretches somewhere around 0.25-0.5mm when tightened (an easily measured difference in length) though the axle will change far less.

Of course, we can all then debate what is 'giving' and how, not that it really matters for the purpose of adjusting bearings. My intuition says that the axle likely does bow internal to the hub where it is unsupported given the length to diameter (aspect) ratio. That could be calculated fairly simply.

HillRider

Which is why aluminum or Ti shafts, particularly in external cam skewers, have less compressive effect that steel shaft skewers.

The claim that: "The force required to compress a steel axle -- albeit follow(sic) -- would be much much greater than a QR could apply with human hands closing it." is also wrong. The cam effect of a good internal cam skewer hugely multiplies the force required to close the lever.

joejack951

First, are there aluminum shaft QR skewers? I've never seen one, just aluminum thru axles.

Second, a Ti shaft can apply just as much compression force as a steel shaft QR, but it needs to be stretched twice as much (which it can handle as long the cam mechanism can provide it, and most external cam QRs can). Similarly, a well-designed external cam skewer can apply as much compression force as an internal cam, it just needs a higher load to close it.

Rowan

If you are going to take me to task, tell us how much force is applied. "Hugely" does not qualify. Then tell us how much compression on the tube that is the axle that will result. Please.

I still say Park has it over all the experts in this forum. They employ people who know their stuff to design tools, and undoubtedly have the knowledge to state that the axles bow rather than compress.

A little test for those who doubt this. Take a plastic tube -- a piece of electrical conduit will do -- and press on the ends of it with your hands. Unless the force is applied exactly to the outer rim, I can (almost) guarantee the tube will bow and bend to collapse. Compression will be zip.

When a QR is applied, it's also not applied directly to the axle, but through the dropouts, and those dropouts are not usually parallel as the force is applied. The result is a bowing of the axle.

HillRider

Yes, I think you are correct that Ti has been used but not Al for skewer shafts.

Also, yes, Ti shafts and an external cams can, theoretically, apply the same force as a steel shaft and internal cam but nearly no one will ever use that much effort to close them. There is a reason frames with horizontal dropouts and/or disc brake forks using qr hubs should always use internal cam steel skewers. We have had numerous threads here reporting slipping rear wheels and disc brake misalignment that were cured by changing the skewer to a steel internal cam type.

Schwinn me

Another armchair engineer checking in.

I know something about nuts and bolts and can confidently say that most of the threads on a regular sized nut do not bear the load under torque, only a few do. A cone can be compared in thread count to a full size nut. I bought some replacement QR Cromo axles, real cheap ones and more expensive vintage. I was curious about the hardness of the replacements compared to the bent and thread damaged originals. My wife is a QC inspector in a machine shop/manufacturing plant. I had her do some Rockwell hardness tests on the axles. Some were harder than the original, others were somewhat softer. There was a fairly wide range in hardness with the originals coming in near the middle in the hardness test. So in my opinion thread contact areas is where most of the end play is being lost. Probably in both thread slack and deflection of the threads on the nuts and axles, softer axles would compound the problem Further, the jamb nut on a cone will not load the two against each other to simulate the same load effect as pressure applied on the outside of both jamb nuts like a QR or axle nuts would do. More opportunity for lost end play or overly tight axle bearings. Since we're only talking a few thousandths here this makes logical sense.

Another area of potential overly tight wheel bearings would come from the warming of alloy hubs and the expansion associated with that.

I'll stick with the Schwinn suggested hint of end play on assembly.

Spoonrobot 

uh...just demonstrate it?

go ahead and tighten the cones/locknuts and leave a little bit of play. Install the wheel, leave the QR loose and give it a spin.

Then tighten and do the same.

Dude most likely thought it was nonsense cause he's never rebuilt a hub. Remember; most bike mechanics are there because they like bikes, not because they're good mechanics or have an intuitive knowledge of mechanical systems.

Schwinn me

I would doubt soundness of wrenching advice from anyone that needs a LBS.

fietsbob

BITD pro team mechanics had a set of dropouts identical to what the team's bikes used.

they used them to adjust every set of spare wheels used in race support,

so the bearing adjustment was right, and the QR itself, were ready to go..




...

corrado33

This x1000. Most bike mechanics are young guys who like to RIDE bikes and thought "Hey! I may as well work at a bike store so I can get parts cheaper."

I can almost guarantee you that half of them have never adjusted a hub. (That type of adjustment would be saved for the "head" mechanic... aka the only one who knows what he's doing.)

I have personally experienced the tightening of the bearings upon installation. I had a QR bike that I adjusted the bearings PERFECTLY off of the bike. I then put the wheel on the bike, clamped down on the QR and... WTF? The wheel would barely spin more than 2 turns or so when I spun it. Yeah yeah yeah unloaded spin tests are not indicative of anything, but I still found this weird. So I remove the wheel, loosen the bearing preload just a touch, put it back on the bike and it spun much longer, indicating that the bearings were, in fact, binding (and it wasn't just seals and stuff.)

You guys can sit here and discuss it all you want, but the plain fact is that it only takes 5 minutes or so to go test it for yourself.

wphamilton

How many turns, times the thread width per turn, is the amount of compression and stretching. I'd say this is true for any nut and bolt, no matter what you're bolting on.

Measure the bolt or QR, before and after tightening. Subtract that from however far the nut tightened. What's left is how much the axle compressed. Makes sense?

IrishBrewer

Engineer here. The formula for axial deflection under load is Delta = (P*L) / (A*E)

P = axial load
L = length of element being compressed (the axle)
A = cross sectional area of element (the axle)
E = modulus of elasticity of the element (the material property of the axle, i.e, steel or aluminum - this is a constant)

Note that the fork or dropouts take some of the compressive load so you are just looking at what load is transferred to the axle after the dropouts/fork make contact.

IrishBrewer

Agreed, this is the best way to assess the effect, a no brainer.

RubeRad

Park Tools cup&cone adjustment page says:

79pmooney

Sorry, Park, I just don't believe that. The skewer is very close to dead center in the axle. How does it exert any off-center bending force? This effect of the bearings being tightened can be see with the bike in as stand or as Cliff notes, by just using a skewer and spacers.

But let's look at simple math and use very basic material laws. Elongation (e) = (force X length) / (area X modulus of elasticity. Again, this is fall of sophomore year, week 4, not high end stuff. For a 9mm axle with as 5.5mm hole for the skewer, the area = 0.062 in^2. The length that is compressed is (roughly) 2 3/4" between bearing races. Steel has a modulus of elasticity of 30,000,000. Now the force applied iai s little harder to come up with but here is a try: Say we apply 10# over 1" (25mm) to the QR lever after all contact is solidly made. Say over that lever travel the cam of the QR extends out an additional 1mm. So the force will be 10 X 25mm/1mm = 250 pounds. (I suspect a good steel QR set tight does better than that, but we will stay here.) So, 250 pounds / 0.062 in^2 = 4032 psi stress in the steel. Not huge but quite real. Now, by that sophomore formula:

e = (250 X 2.75) / (0.062 X 30,000,000) = .0004" = 0.4 mil. Not a lot but any good calipers can measure it. (And this assuming my calc of 250 pounds with 10 pounds on the QR lever. If we use 25 pounds (my levers leave a red impression in my palm) and say only 1/2 mm cam movement, that force would be 1250 pounds and the compression between bearing races 2 mil. That certainly would be the difference between a little wiggle in th e bearings and just right.

So just doing this simple math using very basic engineering principles, compression of the axle to enough to affect the bearings looks completely reasonable. I could go out to the garage and measure the QR dimensions, etc, but it hardly seems worth the effort.

I hope Park has been misquoted, because that idea that the axle is bending sounds even more far fetched after running these numbers. (And just for fun - 1250 pounds of force. 1 mm bending distance. (I don't see how you can get the QR force further that that off the axle centerline.) Bending moment = 1250 X 1mm/25in/mm = 50 inch-pounds. OK, if this moment were instead from an 80 pound weight hung mid-axle (a ridiculous extreme case since there is nothing touching the axle anywhere near the midpoint) the deflection would be 0.001" or 1 mil. The ends (ie the bearing races) would shorten up a lot less. This is getting into more math (parabolas and lengths of vs end-to-end distances and not worth my time, but clearly we are talking far less than the the .4 to 2 mils I came up with above on straight compression.)

I haven't checked this and won't bet my lifer on it. Good thing is that I know from many years of riding with good QRs, I don't have to.

Edit: Park could mean by "flex", straight compression. If so, that is not very elegant wording.

Ben

Reynolds 

I doubt the axle flexes when the QR is tightened. If so cone wear would be uneven, and you usually don't see that on high quality hubs even after a lot of kms.

Bandera

Yep.

-Bandera
Ex-Schwinn Factory Certified Mechanic

joejack951

The bowing/buckling would/could all be happening to the unsupported axle between the bearings. The 'fixed' ends of the axle (where it is supported by the bearings) could remain perfectly square with all the distortion happening elsewhere.

Reynolds 

But then the center part of the axle would flex in a Z-shape? How else could the center flex and the ends remain square?

joejack951

Something like the 'fixed' example:



From: https://en.wikipedia.org/wiki/Buckling

Tape2012

You might be surprised that there are many people like me on this forum who are engineers who design and deal with equipment much more critical than a little slop in a bicycle hub. In my first life I designed coil and leaf springs for General Motors, then at another point designed carbon fiber composites for the aerospace industry. Plus there appear to be several people here who have direct applicable experience and knowledge.

I've dealt with many good mechs/techs who could perform the required tasks but did not understand the underlying principles involved, your Park Tool friend appears to be in that camp. You as well.

If you want a written thesis as proof of something that has had a simple work-around forever, I'm certainly not going to set up any text fixtures or do any extensive calcs. I don't come here to work, you'll just have to take my word. Especially when you come here, ask a question, get the correct answer several times then insult the people who helped you. You must be a blast at parties.

u235

Exactly. Seen the same on sealed and unsealed on multiple wheels and bikes for years even on axels with nuts and not QR. I am now wondering how more people haven't noticed that before.

joejack951

FWIW, I've had no issues other than the occasional less-than-perfect alignment during installation using an external cam quick release with a disc brake (front only) on my cross/commuter bike. Never once slipped but I do make sure to install it very tight every time.

Metaluna

To be fair, most people here are discussing WHY it happens, not if it happens. Other than the idiot shop mechanic Robert P talked to, who should not be allowed to touch hubs. Or at least should stick to cartridge bearings.